Computers, such as personal computers, often include one or more input devices, such as a keyboard and a mouse, that allow a user to control the computer. More sophisticated input devices include voice-recognition input systems and “pen” or stylus based input systems. With pen-based input systems, the user relies on his or her own handwriting or drawing to control or operate the computer. These input systems typically include a hardware device called a “tablet” that is connected to the serial port of the computer. The tablet may include an integrated display screen so that the tablet can serve as both an input and an output device. When operating as an input device, the tablet senses the position of the tip of the pen as it moves across the tablet surface and provides this information to the computer's central processing unit. To provide the user with visual feedback as the pen moves, the computer typically displays “ink” (i.e., a path of pixels tracing the pen's movement) simulating the ink dropped by a real pen. If the tablet has an integrated display screen, this electronic ink is typically drawn directly beneath the tip of the moving pen. For an opaque, input-only tablet, the ink is typically drawn on a normal computer screen to which the tablet has been “mapped”. Whether integrated with the tablet or not, the screen typically displays standard computer-generated information, such as text, images, icons and so forth.
In addition to the tablet, pen-based computers also have a software pen driver that interfaces with the tablet and periodically samples the position of the pen, e.g., 100 times a second. The pen driver passes this ink data to an ink manager that organizes it and coordinates the recognition process. Specifically, the ink manager organizes the ink data into ink strokes, which are defined as the ink data collected until the pen is lifted from the tablet. The ink manager passes the ink strokes to a recognizer that employs various tools, such as neural networks, vocabularies, grammars, etc., to translate the ink strokes into alpha-numeric characters, symbols or shapes. The recognizer may generate several hypotheses of what the ink strokes might be, and each hypothesis may have a corresponding probability. The hypotheses are then provided to an application program which may treat the recognized ink data as an input event.
With Pen Services for Windows 95® from Microsoft Corporation, ink strokes are organized into pen input sessions. See Programmer's Guide to Pen Services for Microsoft Windows 95 (1995 Microsoft Press). A pen input session begins as soon as the user touches the pen to the tablet and ends when the user taps the pen outside of the writing area (e.g., tapping an OK button) or a brief period of inactivity elapses. A new session begins when the user resumes writing. All of the ink strokes corresponding to a given pen input session are accumulated into a single pen data object. An application program associated with the pen input session can basically choose one of two modes of operation. First, the application can choose only to receive the recognition results, thereby allowing the system to process and organize the ink data based on its default settings and to interface with a default recognizer. Alternatively, the application program can request the “raw” ink data and process it in any number of ways. For example, it can buffer the data to delay recognition or it can throw the data away. This ink data is provided to the application on a stroke-by-stroke basis. If the application wishes to have the data recognized, it passes the raw ink data to the recognizer itself. Any special recognition requirements, such as field-specific recognition contexts (e.g., name, address, social security, or other types of input fields), and any unique affiliations between ink input sessions and specific input fields must also be determined on a stroke-by-stroke basis, typically based upon the first stroke that is received.
This collection and organization of ink data into pen input sessions on such a stroke-by-stroke basis has several disadvantages. As noted, to associate ink data with a particular data entry field, the system generally relies solely upon the location of the first ink stroke entered by the user. If a pen input area is declared and subsequent strokes extend outside of that area, the system will not associate those strokes with the data entry field, even though the user may have intended these subsequent ink strokes to be a part of that data entry field. Further, if that first stroke is only slightly misplaced (e.g., if the cross-bar of a capital “T” is drawn first and written too high), the entire subsequent session may be affiliated with the wrong input field. Associating ink with the wrong input field may result in the recognition results flowing to the wrong place, and, if a special recognition context is used for each input field (e.g., name vs. address vs. social security fields, etc.), the wrong context may be applied during recognition.
Even systems that attempt to improve this situation by using each stroke to determine the input field anew, such as the Apple Newton from Apple Computer Inc. of Cupertino, Calif., can suffer from failure modes that make the situation difficult both for end users and for application developers. For example, a word that accidentally spans two input fields even a tiny amount (due, for instance, to a stray ascender, descender, crossbar, or dot) may be broken up into multiple sessions, causing misrecognition and invalid data entries that must be manually corrected.
Accordingly, a need exists for improving the way in which ink data is organized so as to facilitate the recognition process and also to improve the association of ink data to particular data entry fields.